We built our own PIB scoring engine, then tested it honestly against every Current Affairs question on UPSC Prelims 2021 through 2026. For each question with a verifiable PIB source, we asked one thing: would our engine have ranked that exact release in the top 10 of its publishing week?
Of 159 total Current Affairs questions across six papers, 86 had a verifiable English-language PIB release as their source. The remaining 73 were either international news, RBI / SEBI / ministry-website releases, or non-English PIB editions our pipeline could not score. We are honest about this gap — see Limitations below.
A typical PIB publishing week contains about 72 English releases. If you picked 10 of those at random and hoped the UPSC source release was among them, your hit rate would average 10 ÷ 72 ≈ 14%. The exact, basket-weighted random baseline is 17.8% at K=10.
Our engine scores 55.8% at K=10 — three times better than chance. The lift is even larger at the top of the ranking, where attention is scarce.
| Top-K bucket | Random baseline | pibtracker engine | Lift |
|---|---|---|---|
| Top 5 | 8.9% | 43.0% | 4.8× |
| Top 10 | 17.8% | 55.8% | 3.1× |
| Top 20 | 35.6% | 70.9% | 2.0× |
Random baseline is the average of min(K, basket_size) / basket_size across all 86 backtested questions, reflecting the actual basket size each week. Lift = pibtracker ÷ random.
The instinct came first. Aditya Tiwari Sir has been grading Prelims-style mocks since 2018, and the same complaint kept surfacing in batch reviews — students were reading PIB releases nobody would ever ask about, and missing the ones that mattered. We started keeping a quiet list. Not of topics, but of which kind of release tended to surface in exam halls eighteen months later. The exact contour of that list took years to settle. Umang joined in early 2025 and the next four months were mostly arguments about edge cases.
When we sat down to test the instinct formally, the exercise was simple to describe and tedious to execute. We pulled every Current Affairs question from UPSC Prelims 2021 through 2026 — 159 of them. For each one, we hunted down the PIB release the question was almost certainly drawn from. Eighty-six had a clean English PIB anchor we could verify by clicking. The other seventy-three came from outside PIB — international press, RBI circulars, Supreme Court judgments, ministry-website notifications — and are not part of what our filter is built to catch.
Then the test. For each source release we took the surrounding week of PIB output — what a subscriber would have seen on the morning the release went up, nothing more, nothing less. We ran our scoring against that week as it stood at the time, with no knowledge of what UPSC would ask eighteen months later. If our score put the right release near the top of the pile, we counted it a hit. If it buried the release below twenty, we counted it a miss. No double-counting, no per-year curve-fitting, no question-text leakage.
The result is the table above and the per-year pages below. The way our score is built — what makes a release rise, what makes it sink — is the product. We do not publish it. What we do publish is every release we score, every week, alongside its source on the official PIB site, so any aspirant can read both side by side, decide whether our judgment looks sound, and email Aditya Tiwari Sir the moment it does not.
The Union Cabinet's 4 January 2023 approval of the National Green Hydrogen Mission was a single PIB release. It went on to power three Current Affairs questions across two Prelims papers:
The week of 4 January 2023 had 38 English PIB releases. Our engine ranked PRID 1888547 at rank #3 in that week. A subscriber reading the Top 10 that night would have caught it three times over — once for each future Prelims question.
Want more worked examples? Open any year card below — every question we backtested shows its rank, basket size, and a direct link to its PIB source. You can verify any single row yourself in under a minute.
Click any year to open the full question-by-question table for that paper. Every PIB URL is clickable. Every rank is reproducible from the dataset linked at the top of this page.
Two PDFs let any aspirant audit this page without us having to hand over the engine itself.
The Backtest Summary is the document we would hand to a teacher we respect and ask: is this fair? Does the framing hold up? Six pages — the numbers, the worked example, the limitations, nothing else.
The Verification Sheet is the long-form audit trail. Every Current Affairs question from the last six Prelims papers, year by year, with its PIB source release one click away on the official pib.gov.in site. If a row looks wrong, write to us — we update in public.
The underlying mapping work — the depth notes, the cases we argued over, the source releases we eventually rejected — sits in our internal review files. We share it on request with universities, journalists, and serious researchers who write in. There is no download button on purpose. If the methodology matters to you enough that you want the raw material, we want to know who you are.
Found a mapping you disagree with? Write to [email protected] with the question number and the release you believe is the correct source.